Skip to main content
SpringerLink
Log in

Cash flow volatility and corporate bond yield spreads

  • Original Research
  • Published:
Review of Quantitative Finance and Accounting Aims and scope Submit manuscript

Abstract

We conduct a detailed empirical study of the effects of cash flow volatility on corporate bond yield spreads. We use both forward-looking and historical cash flow volatility measures. Using a large sample of transaction prices for investment grade straight bonds, we show that cash flow risk has strong statistical significance and economic effects on spreads, after controlling for a battery of factors which are known to be important determinants of spreads. The effects of cash flow risk are more pronounced for firms that are at greater risk of default, and when cash flow risk is measured based on more recent information. Our results provide empirical support to structural models of bond pricing and emphasize the effect of fundamentals-related information uncertainty on bond prices.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2

Similar content being viewed by others

Notes

  1. The Fitch rating report is available at http://www.reuters.com/article/2014/02/18/fitch-affirms-wyndhams-idr-at-bbb-outloo-idUSFit69088320140218.

  2. A commonly employed proxy for asset volatility is equity return volatility (e.g. Campbell and Taksler 2003; Eom et al. 2004). While equity volatility provides a useful proxy, it may not fully capture asset volatility since, fundamentally, asset values reflect discounted cash flows to the firm as opposed to cash flows to equity.

  3. Formal treatments of the relationship between cash flow volatility and asset return volatility can be found in Goldstein and Zapatero (1996) and Goldstein et al. (2001), and Chapter 11 of Duffie (2001). Under common assumptions, the proportional volatility parameters for asset returns and cash flows are identical.

  4. Some researchers (e.g. Tang and Yan 2010) argue that CDS spreads are a more direct measure of credit risk than corporate bond yield spreads since they are less prone to liquidity and tax effects. However, Nashikkar et al. (2011) provide evidence that CDS spreads do not fully capture credit risk, owing to frictions in arbitrage between CDS and bond markets. Moreover, CDS spreads may also reflect counterparty default risk. In our paper, we focus on corporate bond yield spreads.

  5. An alternative would be to use the Trade Reporting and Compliance Engine (TRACE) data; see, for example, Dick-Nielsen (2009) and Dick-Nielsen et al. (2012). This has the advantage of including all transactions, not just those by insurers. However, TRACE data only started being collected in 2002, with full coverage from 2005. This would leave us with very few quarterly observations, making it difficult to identify different economic regimes or to obtain a time series of standard deviations to characterize time-varying volatilities.

  6. For more details about the FISD and NAIC databases, see Campbell and Taksler (2003).

  7. Merging between FISD and Compustat is not straightforward. The difficulty is that a unique correspondence in firm identifiers between FISD and Compustat is not provided. We carefully address this issue in two steps. We first find all possible matches between these two datasets by an issuer’s CUSIP number. One of the known problems for CUSIP numbers is that over time the same CUSIP number may refer to different firms. To address this problem, we manually sort through each entry by checking the company name in FISD against the company name in Compustat and then eliminate those entries whose names do not match.

  8. Some papers that use TRACE data also adopt similar data filtering procedures (e.g., Dick-Nielsen et al. 2012).

  9. Campbell and Taksler (2003) eliminate AAA (or Aaa) rated bonds because the NAIC data for these issues appears to be somewhat problematic for the early years of the sample, with average yields for these bonds exceeding those of lower rated bonds. Non-investment grade bonds are eliminated because insurance companies typically do not hold such bonds. In fact, Ellul et al. (2011) have shown that regulatory constraints on insurance companies can lead to “fire sales” of bonds that have been downgraded below investment grade, with significant effects on bond prices. The restriction to a minimum of 2 years left until maturity is not explicitly discussed by Campbell and Taksler; however, it is implicit in the values reported in Table 1 of their paper.

  10. Our results are robust to using all trades within the month instead of averaging the within-month trades.

  11. Although the truncation seems arbitrary, our results are qualitatively unchanged if the raw data are used, of if we change filter (ii) to \(-\)20 basis points.

  12. In addition, as shown below in Fig. 1, yield spreads were relatively high throughout 1998–2002. Since Campbell and Taksler’s sample excluded the 2000–2002 period, this is another reason why our mean spreads are somewhat larger than theirs.

  13. The only exception is that BBB rated short-term bonds have higher spreads than BBB rated medium term bonds. This is perhaps due to the issuer quality effect identified by Helwege and Turner (1999): among firms with the same credit rating, riskier firms tend to issue debt with shorter maturity. While Helwege and Turner’s analysis is for non-investment grade issuers, it is possible that this may also happen for relatively low rated investment grade firms.

  14. One possible alternative to the regime-switching approach is to use a GARCH specification. We experimented with replacing our expected cash flow volatility measure with volatility measures based on a GARCH(1,1) model, and found very similar results to those presented here that are based on regime-switching.

  15. Firms began reporting cash flow from operations in fiscal year 1988, as required by the SEC. Allowing for a 1-year adaptation period, we estimate the regime-switching process during fiscal years 1989–2009. We constrain the estimation to firms with at least 40 quarterly cash flow observations to ensure sufficient degrees of freedom in the estimation process.

  16. For each firm, the RCM measure of two-regimes is defined as \(400 \times \frac{1}{T} \sum _{t=1}^T p_t (1-p_t)\), where \(T\) is the number of observations, and \(p_t\) is the smoothed probability of state 1 at time \(t\) (see, e.g., Kim 1994). RCM has a value ranging from 0 to 100. A value of 0 indicates perfect regime classification and a value of 100 implies that no information about the regimes is revealed.

  17. For further details, see Kim (1994) and Hamilton (1994).

  18. We restrict our analysis to firms with at least 6 observations in the past 12 quarters in order to calculate the standard deviation of the current quarter. Removing this restriction does not affect the results.

  19. We thank an anonymous referee for suggesting this measure. To minimize the effect of overlapping observations, we only take observations that are 1-year apart. Therefore, each rolling Std(CF/Debt outflow) estimate uses four observations (i.e. observations in quarters 0, \(-\)4, \(-\)8, and \(-\)12).

  20. About 5 % of our sample firms have \(\beta\) values falling outside this range.

  21. In contrast, the average AR(1) coefficient for market firm value is 0.91 in our sample. For book assets or book equity, the AR(1) coefficient is even higher. As a further comparison, the AR(1) coefficient of firm cash flows is just 0.17, pointing to the high volatility of cash flows.

  22. For example, Bao et al. (2011) measure liquidity using the autocovariance of relative price changes using TRACE data from 2003 through 2009. In order to calculate this measure, a bond must trade on at least \(3/4\) of business days during the sample period for which it is outstanding. By contrast, the mean number of trades per month across all our sample bonds is \(<\)4.

  23. In our sample, the correlation between the logarithm of bond issue size and the logarithm of the market value of equity is 0.43.

  24. Some models imply that the relationship between spread and maturity is nonlinear. For example, the Longstaff and Schwartz (1995) model predicts a hump-shaped effect of maturity on spread. Including a squared log-years to maturity term in our regressions barely changes the magnitude and significance of our results.

  25. A potential issue with these results is that credit rating may be partly capturing the effects of several of the independent variables. To assess this, we ran the regressions without credit rating as a control. The coefficient magnitude and statistical significance for cash flow volatility, equity return volatility, interest coverage, and bond size all increased somewhat under this alternative specification.

  26. If a regressor \(x\) is normally distributed, replacing \(x\) with its standardized counterpart \([x - \text {mean}(x)]/\text {std}(x)\) in the regression results in a new coefficient estimate that equals the original estimate \(\text {est}(x)\) multiplied by \(\text {std}(x)\), without changing its statistical significance. This standardization, however, allows for cross-variable comparisons of significance. Based on this, it is common to measure economic significance of a variable in terms of a one standard deviation change in that variable, i.e. \(\text {est}(x) \times \text {std}(x)\).

  27. The number of firm-year observations in Table 6 is about 10 percent larger than in Table 5 because we are no longer restricted to firms with 40 observations of quarterly cash flows between 1989 and 2009.

  28. In order to save space, Table 8 (and some of the remaining tables below) present results for expected cash flow volatility and one of the historical measures, that being Std(CF/Firm value). The results for the other historical measures are generally similar.

  29. We require at least 15 observations in order to calculate the partial correlation for each issuer. We focus on expected cash flow volatility since it provides a point estimate of volatility that does not use overlapping cash flow observations. The historical volatility measures such as Std(CF/Firm value) use overlapping observations, so it is unclear as to which cash flows in time drive the correlation with spread.

  30. The size of the decline in the number of firm-months for the financial crisis period is puzzling: according to statistics from the Securities Industry and Financial Markets Association, average daily trading volume in corporate bonds in the U.S. was about $17.3 billion from 2002–2006 and about $15 billion from 2007–2009 (see http://www.sifma.org/uploadedFiles/Research/Statistics/StatisticsFiles/CM-US-Bond-Market-Trading-Volume-SIFMA.xls). It is possible that insurance companies reduced their trading activities in corporate bonds during the crisis to a much larger extent than did other participants in these markets. Alternatively, there may be some problems with the NAIC data source during this period. Regardless, it is not clear how representative our sample is during the crisis period, and our inferences about this period are subject to this caveat.

  31. It is worth noting that while the loading of equity return volatility was sharply higher during the financial crisis as compared to the 2002–2006 period, its loading was also much higher during the 2002–2006 period as compared to the 1994–2001 time frame. The importance of this variable was first pointed out by Campbell and Taksler (2003), based on a sample period from 1995–1999. Equity return volatility has thus become even more significant as an explanatory factor for corporate bond yields than was apparent in Campbell and Taksler (2003).

  32. Analyst forecasts are obtained from the Institutional Brokers’ Estimate System (I/B/E/S).

  33. If we repeat the regressions, but drop our cash flow volatility measures, the coefficient estimate, statistical significance, and economic importance of accounting earnings volatility do not change appreciably—in fact, the coefficient estimate and economic significance decline slightly.

References

  • Acharya VV, Carpenter JN (2002) Corporate bond valuation and hedging with stochastic interest rates and endogenous bankruptcy. Rev Finance Stud 15:1355–1383

    Article  Google Scholar 

  • Ang A, Bekaert G (2002) Regime switches in interest rates. J Bus Econ Stat 20:163–182

    Article  Google Scholar 

  • Avramov D, Jostova G, Philipov A (2007) Understanding changes in corporate credit spreads. Finance Anal J 63(2):90–105

    Article  Google Scholar 

  • Bao J, Pan J, Wang J (2011) The illiquidity of corporate bonds. J Finance 66:911–946

    Article  Google Scholar 

  • Benartzi S (2001) Excessive extrapolation and the allocation of 401(k) accounts to company stock. J Finance 56:1747–1764

    Article  Google Scholar 

  • Bernard VL, Thomas JK (1990) Evidence that stock prices do not fully reflect the implications of current earnings for future earnings. J Acc Econ 13:305–340

    Article  Google Scholar 

  • Calvet LE, Fisher AJ (2004) How to forecast long-run volatility: regime switching and the estimation of multifractal processes. J Financ Econom 2:49–83

    Article  Google Scholar 

  • Campbell JY, Taksler GE (2003) Equity volatility and corporate bond yields. J Finance 58:2321–2350

    Article  Google Scholar 

  • Chan K, Jegadeesh N (2004) Market-based evaluation for models to predict bond ratings. Rev Pac Basin Financ Mark Polic 7:153–172

    Article  Google Scholar 

  • Chen L, Lesmond DA, Wei J (2007) Corporate yield spreads and bond liquidity. J Finance 62:119–149

    Article  Google Scholar 

  • Chen L, Collin-Dufresne P, Goldstein RS (2009) On the relation between the credit spread puzzle and the equity premium puzzle. Rev Financ Stud 22:3367–3409

    Article  Google Scholar 

  • Chen P, Wang J, Wu C (2011a) The informativeness of corporate bond trades. Rev Pac Basin Financ Mark Polic 14:367–428

    Article  Google Scholar 

  • Chen RR, Yang TLT (2002) A simple multi-factor, time-dependent-parameter model for the term structure of interest rates. Rev Quant Finance Acc 19:5–20

    Article  Google Scholar 

  • Chen TK, Chen YS, Liao HH (2011b) Labor unions, bargaining power and corporate bond yield spreads: structural credit model perspectives. J Bank Finance 35:2084–2098

    Article  Google Scholar 

  • Chen TK, Liao HH, Huang HC (2013) Macroeconomic risks of supply chain counterparties and corporate bond yield spreads. Rev Quant Finance Acc 40

  • Choi JJ, Laibson D, Madrian BC (2010) Why does the law of one price fail? An experiment on index mutual funds. Rev Financ Stud 23:1405–1432

    Article  Google Scholar 

  • Collin-Dufresne P, Goldstein RS (2001) Do credit spreads reflect stationary leverage ratios? J Finance 56:1929–1957

    Article  Google Scholar 

  • Crabtree AD, Maher JJ (2005) Earnings predictability, bond ratings, and bond yields. Rev Quant Finance Acc 25:233–253

    Article  Google Scholar 

  • Cremers M, Driessen J, Maenhout P, Weinbaum D (2008) Individual stock-option prices and credit spreads. J Bank Finance 32:2706–2715

    Article  Google Scholar 

  • Dick-Nielsen J (2009) Liquidity biases in trace. J Fixed Income 19:43–55

    Article  Google Scholar 

  • Dick-Nielsen J, Feldhutter P, Lando D (2012) Corporate bond liquidity before and after the onset of the subprime crisis. J Financ Econ 103:471–492

    Article  Google Scholar 

  • Diebold FX, Inoue A (2001) Long memory and regime switching. J Econom 105:27–58

    Article  Google Scholar 

  • Duffie D (2001) Dyn Asset Pricing Theory, 3rd edn. Princeton University Press, Princeton

    Google Scholar 

  • Ellul A, Jotikasthira C, Lundblad CT (2011) Regulatory pressure and fire sales in the corporate bond market. J Financ Econ 101:596–620

    Article  Google Scholar 

  • Elton EJ, Gruber MJ, Agrawal D, Mann C (2001) Explaining the rate spread on corporate bonds. J Finance 56:247–277

    Article  Google Scholar 

  • Eom YH, Helwege J, Huang JZ (2004) Structural models of corporate bond pricing: an empirical analysis. Rev Financ Stud 17:499–544

    Article  Google Scholar 

  • Ericsson J, Jacobs K, Oviedo R (2009) The determinants of credit default swap premia. J Financ Quant Anal 44:109–132

    Article  Google Scholar 

  • Fama E, MacBeth J (1973) Risk, return, and equilibrium: empirical tests. J Polit Econ 81:607–636

    Article  Google Scholar 

  • Goldstein R, Zapatero F (1996) General equilibrium with constant relative risk aversion and Vasicek interest rates. Math Finance 6:331–340

    Article  Google Scholar 

  • Goldstein R, Ju N, Leland H (2001) An EBIT-based model of optimal capital structure. J Bus 74:483–512

    Article  Google Scholar 

  • Gonzalez MdlO, Skinner FS, Agyei-Ampomah S (2013) Term structure information and bond strategies. Rev Quant Financ Acc 41:53–74

    Article  Google Scholar 

  • Gopalan R, Song F, Yerramilli V (2013) Debt maturity structure and credit quality. J Financ Quant Anal (forthcoming)

  • Graham JR, Harvey CR (2001) The theory and practice of corporate finance: evidence from the field. J Financ Econ 60:187–243

    Article  Google Scholar 

  • Gray S (1996) Modeling the conditional distribution of interest rates as a regime-switching process. J Financ Econ 42:27–62

    Article  Google Scholar 

  • Güntay L, Hackbarth D (2010) Corporate bond credit spreads and forecast dispersion. J Bank Finance 34:2328–2345

    Article  Google Scholar 

  • Hamilton JD (1989) A new approach to the economic analysis of nonstationary time series and the business cycle. Econometrica 57:357–384

    Article  Google Scholar 

  • Hamilton JD (1990) Analysis of time series subject to change in regime. J Econom 45:39–70

    Article  Google Scholar 

  • Hamilton JD (1994) Time series analysis. Princeton University Press, Princeton

    Google Scholar 

  • Hansen B (1992) The likelihood ratio test under non-standard conditions: testing the Markov-switching model of GNP. J Appl Econom 7:561–582

    Article  Google Scholar 

  • Helwege J, Turner CM (1999) The slope of the credit yield curve for speculative-grade issuers. J Finance 54:1869–1884

    Article  Google Scholar 

  • Kim CJ (1994) Dynamic linear models with Markov-switching. J Econom 60:1–22

    Article  Google Scholar 

  • Leland HE, Toft KB (1996) Optimal capital structure, endogenous bankruptcy, and the term structure of credit spreads. J Finance 51:987–1019

    Article  Google Scholar 

  • Lin H, Wang J, Wu C (2011) Liquidity risk and expected corporate bond returns. J Financ Econ 99:628–650

    Article  Google Scholar 

  • Longstaff FA, Schwartz ES (1995) A simple approach to valuing risky fixed and floating rate debt. J Finance 50:789–819

    Article  Google Scholar 

  • Lu CW, Chen TK, Liao HH (2010) Information uncertainty, information asymmetry, and corporate bond yield spreads. J Bank Finance 34:2265–2279

    Article  Google Scholar 

  • Mansi SA, Maxwell WF, Miller DP (2011) Analyst forecast characteristics and the cost of debt. Rev Acc Stud 16:116–142

    Article  Google Scholar 

  • Merton RC (1974) On the pricing of corporate debt: the risk structure of interest rates. J Finance 29:449–470

    Google Scholar 

  • Minton BA, Schrand C (1999) The impact of cash flow volatility on discretionary investment and the costs of debt and equity financing. J Financ Econ 54:423–460

    Article  Google Scholar 

  • Molina CA (2005) Are firms underleveraged? An examination of the effect of leverage on default probabilities. J Finance 60:1427–1459

    Article  Google Scholar 

  • Nashikkar A, Subrahmanyam MG, Mahanti S (2011) Liquidity and arbitrage in the market for credit risk. J Financ Quant Anal 46:627–656

    Article  Google Scholar 

  • Newey WK, West KD (1987) A simple, positive semi-definite, heteroskedasticity and autocorrelation consistent covariance matrix. Econometrica 55:703–708

    Article  Google Scholar 

  • Ofek E, Richardson M (2003) Dotcom mania: the rise and fall of internet stock prices. J Finance 58:1113–1137

    Article  Google Scholar 

  • Tang DY, Yan H (2010) Market conditions, default risk and credit spreads. J Bank Finance 34:724–734

    Article  Google Scholar 

  • Tversky A, Kahneman D (1974) Judgment under uncertainty: heuristics and biases. Science 185:1124–1131

    Article  Google Scholar 

  • Vissing-Jorgensen A (2004) Perspectives on behavioral finance: does “irrationality” disappear with wealth? Evidence from expectations and actions. In: NBER Macroeconomics Annual 2003, NBER Chapters, National Bureau of Economic Research Inc, vol 18, pp 139–208

Download references

Acknowledgments

This research was funded by the Social Sciences and Humanities Research Council of Canada. This paper has benefitted from comments provided by Brian Barrett, Kelly Cai, Jay Huang, Madhu Kalimipalli, and seminar participants at the Canadian Mathematical Society Winter 2007 meetings, the Southern Finance Association 2008 meetings, the Northern Finance Association 2009 meetings, the Eastern Finance Association 2010 meetings, the Midwest Finance Association 2013 meetings, the University of Waterloo, and Wilfrid Laurier University. Darren Henderson and Dian Zhu provided excellent research assistance.

Author information

Authors and Affiliations

  1. School of Accounting and Finance, University of Waterloo, 200 University Avenue West, N2L 3G1, Waterloo, ON, Canada

    Alan V. S. Douglas, Alan G. Huang & Kenneth R. Vetzal

Authors
  1. Alan V. S. Douglas
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Alan G. Huang
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Kenneth R. Vetzal
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Alan G. Huang.

Appendix 1: A regime-switching specification for cash flows

Appendix 1: A regime-switching specification for cash flows

We use a regime-switching specification for the evolution of cash flows (Hamilton 1989). For comparability across firms, cash flows for each firm are scaled by the firm’s total assets. The model incorporates two distinct cash flow volatility regimes. While in principle it is straightforward to use more regimes, for simplicity we restrict attention to the bivariate case. In particular, cash flows for a firm from time \(t-1\) to time \(t\) are characterized by

$$\begin{aligned} CF_t = \mu _k + \sigma _k \epsilon _t, \quad k \in \{1,2\}, \end{aligned}$$

where \(CF_t\) is cash flows generated by the firm relative to total assets during the period, \(\mu _k\) is the expected value of the firm’s cash flows in state \(k,\, \sigma _k\) is the standard deviation of the firm’s cash flows in state \(k\), and \(\epsilon _t\) is a standard normal random variable. For our purposes, the most important feature of this empirical specification is that it allows for different levels of cash flow volatility across the two states. The mean level of cash flows is similarly assumed to vary across states. Of course, both the mean and volatility are presumed constant within a state of the economy. At first glance, this might seem unreasonable since it appears that we are not allowing for growth in cash flows over time. However, recall that in this specification cash flows are scaled by total assets. To the extent that the growth in cash flows is accompanied by similar growth in assets, such effects will cancel out. Shifts between the two possible states are governed by a transition matrix P. Following Hamilton (1989), P is in the form of a first-order Markov chain, with element \(p_{jk}\) defined as the probability of being in state \(j\) at time \(t\) conditional on being in state \(k\) at time \(t-1\), for \(j,k \in \{1,2\}\). We use the maximum likelihood approach described by Hamilton (1994) to estimate the parameters of the regime-switching model, i.e. to estimate \(\mu _{k},\, \sigma _k\), and the transition matrix P. We fit the specification for each firm, treating all firms independently.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Douglas, A.V.S., Huang, A.G. & Vetzal, K.R. Cash flow volatility and corporate bond yield spreads. Rev Quant Finan Acc 46, 417–458 (2016). https://doi.org/10.1007/s11156-014-0474-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11156-014-0474-0

Keywords

JEL Classification

Search

Navigation